Databases Reference
In-Depth Information
The query-by-visual-example provides sketch retrieval facility to find similar
pictorial data with out textual information. The query-by-subjective-description
provides a facility for a user to show his own emotional representations to find
pictorial data which is appropriate to his subjective interpretation automatically
evaluating the content of the pictorial data. Those systems have been implemented
with several functions for computing correlations between the user's request
and retrieval candidate pictorial data.
As one of the database systems dealing with Kansei information, we have
introduced a semantic associative search system for images
5)
. The semantic
associative search system realizes image data retrieval by receiving keywords
representing the user's impression and the image's contents. This system provides
several functions for performing the semantic associative search for images by using
the metadata representing the features of images. These functions are realized by
using the mathematical model of meaning
4,
5)
. The mathematical model of meaning
provides semantic functions for computing specific meanings of keywords which
are used for retrieving images unambiguously and dynamically. The main feature of
this model is that the semantic associative search is performed in the orthogonal
semantic space. This space is created for dynamically computing semantic equivalence
or similarity between the metadata items of the images and keywords.
A structure of Kansei information database has been proposed for supporting
design processes by constructing static and dynamic image information of human
motions and positions
15)
. The main purpose of this project is to structure
information regarding movement and posture of human bodies as databases for
supporting tools for design ideas. The movement of hands(animated, still images)
and operation sound (effects) can be included as the contents of the database.
Currently, images of hand motions operating equipments are stored in databases
and those data are manipulated by the senses of sight, touching, and listening.
A learning mechanism is very important for database systems dealing with Kansei
information to adapt retrieval results according to individual variation and
improving accuracy of the retrieval results. Such database systems might not always
select accurate and appropriate data items from databases, because the judgement
of accuracy for the retrieval results is strongly dependent on individual variation.
In the learning, if inappropriate retrieval results for a request are extracted by the
system, accurate data items which must be the retrieval results are specified as
suggestions. Then, the learning mechanism is applied to the system to extract the
appropriate retrieval results in subsequent requests.
Several approaches for adapting retrieval results to the user's impression have
been proposed. In
11)
, in the framework of query-by-visual-example, individual
variations of users are reflected by adapting individual user's information to
database contents. This method is based on the computations of correlations
between images data and user's individual data which are represented in vectors
which consist of color elements.
In
8)
, the learning mechanism has been proposed for applying it to the semantic
associative search system. In this learning, if inappropriate retrieval results for a
